[BioC] DESeq2 in R_devel: a bug?

[Sylvain Foisy Ph. D.]

Hi,

I used DEseq2 v. 1.3 under R-devel because I have many different values for my design (celltype) and it worked ok. I upgraded my libraries this morning (bad,bad,bad…) and now it is not working. Might this be a bug?

> exp.data<-read.table("./count_table_by_gene.txt",header=T,row.names=1)
> exp.ann<-read.table("./desc_colonnes.txt",header=T)
> genex<-DESeqDataSetFromMatrix(
+ countData=exp.data,
+ colData=exp.ann,
+ design=~ celltype)
> genex
class: DESeqDataSet 
dim: 25192 103 
exptData(0):
assays(1): counts
rownames(25192): A1BG A1BG-AS1 ... ZZEF1 ZZZ3
rowData metadata column names(0):
colnames(103): 169377_LYMPHO_B_accepted_hits.bam
  169377_celltype1_accepted_hits.bam ... 999046_celltype_m_accepted_hits.bam
  999046_celltype_n_accepted_hits.bam
colData names(2): sizeFactor celltype
> genex<-DESeq(genex)
using pre-existing size factors
estimating dispersions
gene-wise dispersion estimates
Error in relevel.factor(coldata[[v]], as.character(coldata[[v]][idx])) : 
  'ref' must be of length one

My count files were created using easyRNASeq-stable.

Thanks in advance

Sylvain